Import libraries¶
In [1]:
import numpy as np
from numpy import zeros
from numpy import ones
from numpy.random import randn
from numpy.random import randint
import matplotlib.pyplot as plt
import pandas as pd
import tensorflow as tf
import keras
from keras import layers
from keras import Input
from keras.models import Sequential, Model
from keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, Dropout,Input, BatchNormalization,Conv2DTranspose,UpSampling2D, LeakyReLU,Reshape
from keras.optimizers import Adam ,RMSprop
import cv2
import os
from tqdm import tqdm
import re
from PIL import Image
from tensorflow.keras.utils import img_to_array
/opt/conda/lib/python3.10/site-packages/scipy/__init__.py:146: UserWarning: A NumPy version >=1.16.5 and <1.23.0 is required for this version of SciPy (detected version 1.23.5
warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"
/opt/conda/lib/python3.10/site-packages/tensorflow_io/python/ops/__init__.py:98: UserWarning: unable to load libtensorflow_io_plugins.so: unable to open file: libtensorflow_io_plugins.so, from paths: ['/opt/conda/lib/python3.10/site-packages/tensorflow_io/python/ops/libtensorflow_io_plugins.so']
caused by: ['/opt/conda/lib/python3.10/site-packages/tensorflow_io/python/ops/libtensorflow_io_plugins.so: undefined symbol: _ZN3tsl6StatusC1EN10tensorflow5error4CodeESt17basic_string_viewIcSt11char_traitsIcEENS_14SourceLocationE']
warnings.warn(f"unable to load libtensorflow_io_plugins.so: {e}")
/opt/conda/lib/python3.10/site-packages/tensorflow_io/python/ops/__init__.py:104: UserWarning: file system plugins are not loaded: unable to open file: libtensorflow_io.so, from paths: ['/opt/conda/lib/python3.10/site-packages/tensorflow_io/python/ops/libtensorflow_io.so']
caused by: ['/opt/conda/lib/python3.10/site-packages/tensorflow_io/python/ops/libtensorflow_io.so: undefined symbol: _ZTVN10tensorflow13GcsFileSystemE']
warnings.warn(f"file system plugins are not loaded: {e}")
Load the data¶
In [2]:
# to get the files in proper order
def sorted_alphanumeric(data):
convert = lambda text: int(text) if text.isdigit() else text.lower()
alphanum_key = lambda key: [convert(c) for c in re.split('([0-9]+)',key)]
return sorted(data,key = alphanum_key)
# Change size of the image
SIZE = 128
image = []
path = '../input/face-mask-lite-dataset/without_mask'
files = os.listdir(path)
files = sorted_alphanumeric(files)
for i in tqdm(files):
if i == 'seed10000.png':
break
else:
img = cv2.imread(path + '/'+i,1)
# convert BGR to RGB
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
#resizing image
img = cv2.resize(img, (SIZE, SIZE))
img = (img - 127.5) / 127.5
imh = img.astype(float)
image.append(img_to_array(img))
100%|██████████| 10000/10000 [10:11<00:00, 16.34it/s]
Data Visualization¶
In [3]:
_,ax = plt.subplots(5,5, figsize = (8,8))
for i in range(5):
for j in range(5):
ax[i,j].imshow(image[5*i+j])
ax[i,j].axis('off')
In [4]:
data = np.array(image)
print(data.shape)
(10000, 128, 128, 3)
In [5]:
batch_size = 32
dataset=tf.data.Dataset.from_tensor_slices(np.array(image)).batch(batch_size)
model Architect¶
In [6]:
#discriminator
def discriminator():
model = tf.keras.models.Sequential()
model.add(tf.keras.layers.Input((128, 128, 3)))
model.add(tf.keras.layers.Conv2D(128,4, strides=2, padding='same',kernel_initializer='he_normal', use_bias=False))
model.add(tf.keras.layers.BatchNormalization())
model.add(tf.keras.layers.LeakyReLU())
model.add(tf.keras.layers.Conv2D(128,4, strides=2, padding='same',kernel_initializer='he_normal', use_bias=False))
model.add(tf.keras.layers.BatchNormalization())
model.add(tf.keras.layers.LeakyReLU())
model.add(tf.keras.layers.Conv2D(256,4, strides=2, padding='same',kernel_initializer='he_normal', use_bias=False))
model.add(tf.keras.layers.BatchNormalization())
model.add(tf.keras.layers.LeakyReLU())
model.add(tf.keras.layers.Conv2D(256,4, strides=2, padding='same',kernel_initializer='he_normal', use_bias=False))
model.add(tf.keras.layers.BatchNormalization())
model.add(tf.keras.layers.LeakyReLU())
model.add(tf.keras.layers.Conv2D(512,4, strides=2, padding='same',kernel_initializer='he_normal', use_bias=False))
model.add(tf.keras.layers.LeakyReLU())
model.add(tf.keras.layers.Flatten())
model.add(tf.keras.layers.Dense(1,activation = 'sigmoid'))
return model
discriminator = discriminator()
print(discriminator.summary())
Model: "sequential"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
conv2d (Conv2D) (None, 64, 64, 128) 6144
batch_normalization (BatchN (None, 64, 64, 128) 512
ormalization)
leaky_re_lu (LeakyReLU) (None, 64, 64, 128) 0
conv2d_1 (Conv2D) (None, 32, 32, 128) 262144
batch_normalization_1 (Batc (None, 32, 32, 128) 512
hNormalization)
leaky_re_lu_1 (LeakyReLU) (None, 32, 32, 128) 0
conv2d_2 (Conv2D) (None, 16, 16, 256) 524288
batch_normalization_2 (Batc (None, 16, 16, 256) 1024
hNormalization)
leaky_re_lu_2 (LeakyReLU) (None, 16, 16, 256) 0
conv2d_3 (Conv2D) (None, 8, 8, 256) 1048576
batch_normalization_3 (Batc (None, 8, 8, 256) 1024
hNormalization)
leaky_re_lu_3 (LeakyReLU) (None, 8, 8, 256) 0
conv2d_4 (Conv2D) (None, 4, 4, 512) 2097152
leaky_re_lu_4 (LeakyReLU) (None, 4, 4, 512) 0
flatten (Flatten) (None, 8192) 0
dense (Dense) (None, 1) 8193
=================================================================
Total params: 3,949,569
Trainable params: 3,948,033
Non-trainable params: 1,536
_________________________________________________________________
None
In [7]:
#generator
latent_dim = 100
def generator():
model = tf.keras.Sequential()
model.add(layers.Dense(128*128*3, use_bias=False, input_shape=(latent_dim,)))
model.add(layers.Reshape((128,128,3)))
model.add(tf.keras.layers.Conv2D(128,4, strides=1, padding='same',kernel_initializer='he_normal', use_bias=False))
model.add(tf.keras.layers.Conv2D(128,4, strides=2, padding='same',kernel_initializer='he_normal', use_bias=False))
model.add(tf.keras.layers.BatchNormalization())
model.add(tf.keras.layers.LeakyReLU())
model.add(tf.keras.layers.Conv2D(256,4, strides=1, padding='same',kernel_initializer='he_normal', use_bias=False))
model.add(tf.keras.layers.Conv2D(256,4, strides=2, padding='same',kernel_initializer='he_normal', use_bias=False))
model.add(tf.keras.layers.BatchNormalization())
model.add(tf.keras.layers.LeakyReLU())
model.add(tf.keras.layers.Conv2DTranspose(512, 4, strides=1,padding='same',kernel_initializer='he_normal',use_bias=False))
model.add(tf.keras.layers.Conv2D(512,4, strides=2, padding='same',kernel_initializer='he_normal', use_bias=False))
model.add(tf.keras.layers.LeakyReLU())
model.add(tf.keras.layers.Conv2DTranspose(512, 4, strides=1,padding='same',kernel_initializer='he_normal',use_bias=False))
model.add(tf.keras.layers.Conv2DTranspose(512, 4, strides=2,padding='same',kernel_initializer='he_normal',use_bias=False))
model.add(tf.keras.layers.BatchNormalization())
model.add(tf.keras.layers.LeakyReLU())
model.add(tf.keras.layers.Conv2DTranspose(256, 4, strides=1,padding='same',kernel_initializer='he_normal',use_bias=False))
model.add(tf.keras.layers.Conv2DTranspose(256, 4, strides=2,padding='same',kernel_initializer='he_normal',use_bias=False))
model.add(tf.keras.layers.BatchNormalization())
model.add(tf.keras.layers.Conv2DTranspose(128, 4, strides=2,padding='same',kernel_initializer='he_normal',use_bias=False))
model.add(tf.keras.layers.Conv2DTranspose(128, 4, strides=1,padding='same',kernel_initializer='he_normal',use_bias=False))
model.add(tf.keras.layers.BatchNormalization())
model.add(tf.keras.layers.Conv2DTranspose(3,4,strides = 1, padding = 'same',activation = 'tanh'))
return model
generator = generator()
print(generator.summary())
Model: "sequential_1"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
dense_1 (Dense) (None, 49152) 4915200
reshape (Reshape) (None, 128, 128, 3) 0
conv2d_5 (Conv2D) (None, 128, 128, 128) 6144
conv2d_6 (Conv2D) (None, 64, 64, 128) 262144
batch_normalization_4 (Batc (None, 64, 64, 128) 512
hNormalization)
leaky_re_lu_5 (LeakyReLU) (None, 64, 64, 128) 0
conv2d_7 (Conv2D) (None, 64, 64, 256) 524288
conv2d_8 (Conv2D) (None, 32, 32, 256) 1048576
batch_normalization_5 (Batc (None, 32, 32, 256) 1024
hNormalization)
leaky_re_lu_6 (LeakyReLU) (None, 32, 32, 256) 0
conv2d_transpose (Conv2DTra (None, 32, 32, 512) 2097152
nspose)
conv2d_9 (Conv2D) (None, 16, 16, 512) 4194304
leaky_re_lu_7 (LeakyReLU) (None, 16, 16, 512) 0
conv2d_transpose_1 (Conv2DT (None, 16, 16, 512) 4194304
ranspose)
conv2d_transpose_2 (Conv2DT (None, 32, 32, 512) 4194304
ranspose)
batch_normalization_6 (Batc (None, 32, 32, 512) 2048
hNormalization)
leaky_re_lu_8 (LeakyReLU) (None, 32, 32, 512) 0
conv2d_transpose_3 (Conv2DT (None, 32, 32, 256) 2097152
ranspose)
conv2d_transpose_4 (Conv2DT (None, 64, 64, 256) 1048576
ranspose)
batch_normalization_7 (Batc (None, 64, 64, 256) 1024
hNormalization)
conv2d_transpose_5 (Conv2DT (None, 128, 128, 128) 524288
ranspose)
conv2d_transpose_6 (Conv2DT (None, 128, 128, 128) 262144
ranspose)
batch_normalization_8 (Batc (None, 128, 128, 128) 512
hNormalization)
conv2d_transpose_7 (Conv2DT (None, 128, 128, 3) 6147
ranspose)
=================================================================
Total params: 25,379,843
Trainable params: 25,377,283
Non-trainable params: 2,560
_________________________________________________________________
None
In [8]:
#before train
noise = np.random.normal(-1,1,(1,100))
img = generator(noise)
plt.imshow(img[0,:,:,0])
plt.show()
In [9]:
# optimizer
optimizer = tf.keras.optimizers.legacy.RMSprop(
learning_rate=.0001,
clipvalue=1.0,
decay=1e-8
)
cross_entropy = tf.keras.losses.BinaryCrossentropy(from_logits = True)
In [10]:
def generator_loss(fake_output):
return cross_entropy(tf.ones_like(fake_output),fake_output)
def discriminator_loss(fake_output, real_output):
fake_loss = cross_entropy(tf.zeros_like(fake_output),fake_output)
real_loss = cross_entropy(tf.ones_like(real_output),real_output)
return fake_loss + real_loss
training model¶
In [11]:
# train
def train_steps(images):
noise = np.random.normal(0,1,(batch_size,latent_dim))
with tf.GradientTape() as gen_tape , tf.GradientTape() as disc_tape:
generated_images = generator(noise)
fake_output = discriminator(generated_images)
real_output = discriminator(images)
gen_loss = generator_loss(fake_output)
dis_loss = discriminator_loss(fake_output, real_output)
gradient_of_generator = gen_tape.gradient(gen_loss, generator.trainable_variables)
gradient_of_discriminator = disc_tape.gradient(dis_loss, discriminator.trainable_variables)
optimizer.apply_gradients(zip(gradient_of_generator,generator.trainable_variables))
optimizer.apply_gradients(zip(gradient_of_discriminator, discriminator.trainable_variables))
loss = {'gen loss':gen_loss,
'disc loss': dis_loss}
return loss
In [12]:
latent_dim = 100
#function to plot generated images
def plot_generated_images(square = 5, epochs = 0):
plt.figure(figsize = (10,10))
for i in range(square * square):
if epochs != 0:
if(i == square //2):
plt.title("Generated Image at Epoch:{}\n".format(epochs), fontsize = 32, color = 'black')
plt.subplot(square, square, i+1)
noise = np.random.normal(0,1,(1,latent_dim))
img = generator(noise)
plt.imshow(np.clip((img[0,...]+1)/2, 0, 1))
plt.xticks([])
plt.yticks([])
plt.grid()
In [13]:
import time
def train(epochs,dataset):
for epoch in range(epochs):
start = time.time()
print("\nEpoch : {}".format(epoch + 1))
for images in dataset:
loss = train_steps(images)
print(" Time:{}".format(np.round(time.time() - start),2))
print("Generator Loss: {} Discriminator Loss: {}".format(loss['gen loss'],loss['disc loss']))
Train model¶
In [14]:
train(20,dataset)
Epoch : 1
/opt/conda/lib/python3.10/site-packages/keras/backend.py:5703: UserWarning: "`binary_crossentropy` received `from_logits=True`, but the `output` argument was produced by a Sigmoid activation and thus does not represent logits. Was this intended? output, from_logits = _get_logits(
Time:290.0 Generator Loss: 1.2232353687286377 Discriminator Loss: 0.7618696689605713 Epoch : 2 Time:281.0 Generator Loss: 0.59926837682724 Discriminator Loss: 1.454757571220398 Epoch : 3 Time:281.0 Generator Loss: 0.5312163233757019 Discriminator Loss: 1.4760253429412842 Epoch : 4 Time:281.0 Generator Loss: 0.6700804233551025 Discriminator Loss: 1.1450787782669067 Epoch : 5 Time:281.0 Generator Loss: 0.6326004862785339 Discriminator Loss: 1.186072587966919 Epoch : 6 Time:281.0 Generator Loss: 0.7092082500457764 Discriminator Loss: 1.170581340789795 Epoch : 7 Time:281.0 Generator Loss: 0.5865207314491272 Discriminator Loss: 1.176328420639038 Epoch : 8 Time:281.0 Generator Loss: 0.5529289245605469 Discriminator Loss: 1.1728383302688599 Epoch : 9 Time:281.0 Generator Loss: 0.5605541467666626 Discriminator Loss: 1.3144148588180542 Epoch : 10 Time:281.0 Generator Loss: 1.0353913307189941 Discriminator Loss: 1.0253568887710571 Epoch : 11 Time:322.0 Generator Loss: 0.48352938890457153 Discriminator Loss: 1.3257793188095093 Epoch : 12 Time:281.0 Generator Loss: 0.7955543994903564 Discriminator Loss: 1.072878360748291 Epoch : 13 Time:281.0 Generator Loss: 0.7340448498725891 Discriminator Loss: 1.0307979583740234 Epoch : 14 Time:281.0 Generator Loss: 0.6468023657798767 Discriminator Loss: 1.0698760747909546 Epoch : 15 Time:322.0 Generator Loss: 0.58468097448349 Discriminator Loss: 1.105216145515442 Epoch : 16 Time:281.0 Generator Loss: 2.3347020149230957 Discriminator Loss: 1.2769235372543335 Epoch : 17 Time:281.0 Generator Loss: 0.9419915676116943 Discriminator Loss: 0.9583885669708252 Epoch : 18 Time:281.0 Generator Loss: 0.633262038230896 Discriminator Loss: 1.035232663154602 Epoch : 19 Time:281.0 Generator Loss: 0.5352416038513184 Discriminator Loss: 1.1685130596160889 Epoch : 20 Time:281.0 Generator Loss: 1.067091464996338 Discriminator Loss: 0.7749885320663452
plot generated image¶
In [15]:
plot_generated_images(1)
In [16]:
plot_generated_images(3)
In [17]:
plot_generated_images(6)
In [18]:
plot_generated_images(17)
